Principal Component Analysis Explained

Principal Component Analysis (PCA) is commonly employed in research to identify patterns. This presentation is a quick summary of PCA, and is intended for the non-statistician.

[marksanda]

The best explanation of PCA that I have ever heard, hands down.

[suhasjadhav]

It is one of the most intuitive and crystal clear explanation of PCA concept in less than 10 minutes.

[j.larrycampbell]

Thank you very much, @RayBiotech, for providing this excellent description of PCA!

[chicanicajfcp]

Wrong! The maximum number of PC's are min(n-1, k). In the example you had 10 samples and 40 attributes. The answer should be <= 9 and usually a lot fewer meaningful components. If you have two samples they can always be represented as a line and with 3 samples a plane with no loss of information, same is valid for higher dimensions. Regardless of the number of samples, the number of PC's can never exceed the original number of variables.